Engineous Software 2008 isight-fd Overview European Symposium Holger Wenzel and Seminar Series Engineous Software GmbH Engineous Software, Inc. ENGINEOUS A Dassault Systèmes SOFTWARE, Company INC 14-15 October April, 9., 2008 Copyright 2008, Dassault Systèmes
2 About isight-fd Desktop workflow builder Parametric, hierarchical, nested Drag and drop environment Powerful design drivers: Multi-Discipline Optimization Design of Experiments Monte Carlo Analysis Approximation Methods Desktop automation Automate simulation process to explore design space Distributed/parallel computing Multi-run Visual Post-processing Understand design trade-offs Surf through the design space in real-time.
3 Functional Overview
4 isight-fd Functionality Visualize Process: Advanced GUI provides rich visual feedback of workflow and design exploration results Connect Process: Modular components make it easy to interface with external programs and post-processing tools Extend Process: Seamlessly expand from desktop to enterprise level for a true collaborative design environment
5 Focus of isight-fd Ease of use Workflow and setup wizards Complex process capture and automation Task plan, nested drivers, Design drivers NLPQL, Pointer, OLH, MOGA, etc. Approximation modeling RSM, RBF neural network, error checking Visualization Engineering data mining, design space surfing Extensibility Can be connected to FIPER ACS and other job managing system.
6 isight-fd Structure Design Gateway Build Workflow Model Runtime Gateway Run Workflow Model Components Design Drivers Visualize Data
7 isight-fd Operation Design Gateway Build Workflow Model Runtime Gateway Run Workflow Model Build process Workflow model Select components Map parameters Establish loop and/or branch controls Select optimization/sampling using Task Plan capability Select graph templates Run Workflow and view/inspect data Monitor run status View/modify parameters Generate history plots Display graphs Post-processing Engineering Data Mining Visual Design Driver
8 Wrapping Components Wrapped software programs that have single or multiple features can be inserted into the isight-fd Workflow. Custom setup GUI is shown to the user Input and output parameters are defined Icon is created for toolbar Program execution is triggered as the workflow is executed Status is monitored Results are stored Setup GUI Example: EXCEL Component Input Parameters Execute Excel Output Parameters ICON Manage Status Post Process Workflow Component
9 Design Gateway User Interface for Building Workflow Models
10 Building a Workflow Model Workflow Components Drag & Drop Cut & Paste Right-click menu Shortcut keys Multi-level Conditional Parallel
11 Show and Map Parameters Show parameters of each simcode component Show parameter mapping between simcodes Upstream output parameters Downstream input parameters
12 Workflow Model Execution Flexibility Can execute an entire workflow, or only individual components Workflow model is easy to build and check by adding components, selecting them for execution, and inspecting the process step-by-step. Execute a complete model Execute with a sub-model Execute a selected component
14 Internationalization (I18N) Environment GUI Menu Component name Parameter name Working directory Save model name On-line Help Supported OS: Windows Linux Solaris AIX HP-UX
15 isight-fd Components Process Component Activity Component Design Driver Component
16 Definition of Components Process Component Contains sub-models Controls sub-models Design Driver Component Optimization Design of Experiments Monte Carlo Simulation Approximation models Activity Component Calculations and simulation codes Set conditions to execute features Execute
17 isight-fd Process Components Process Components Task Loop DOE Optimization Monte Carlo Fundamental process component Execute Sub-workflow Control iteration of sub-workflow For, For each, Do Until Design of Experiments methods Optimization methods Monte Carlo methods
18 isight-fd Activity Components Activity Components used to build automated Workflows: Data Exchanger OS Command Simcode Calculator Excel Word Mail Read and write numerical values and character string data described in a text file Execute commands, batch files, scripts Execute simulation codes and read and write associated input and output files, execute commands Calculate with parameters and functions Read and write cells in Microsoft Excel spreadsheets, execute macros Read and write Microsoft Word documents to generate reports automatically Send e-mail containing results and data generated during execution, attach data files
19 isight-fd Activity Components Activity Components used for various purposes: Database Script COM Pause isight Interface with SQL compliant rational database (Oracle, DB2, Access, MySQL) to store input and output data Execute Java script which is not necessary to be compiled Interface with COM (Component Object Model) objects Pause the workflow during execution with conditional settings. Shows a dialog to determine go or no go. Execute isight description file Approximation Generate, inspect, and visualize approximation models
20 Solution Components Direct Interfaces CAD, CAE, CFD, COST, DB, Microsoft Apps, Etc. Access Excel Excel General Interfaces Calculations, In-House Programs, Scripts, Etc. Calculator Simcode COM Link E-mail Data MatchXML Parser DLL Link Text Parser Pause Script OS Command Custom Interfaces User Developed or Engineous Services Customer Examples: Combat XXI Vensim
21 Solution Components Suite CAE Performance Simulation Abaqus: Exchange data with and execute Abaqus FEA ADAMS: Exchange data with and execute ADAMS/View ADAMS/Car: Exchange data with and execute ADAMS/Car ADAMS/Chassis: Exchange data with and execute ADAMS/Chassis AMESim: Exchange data with AMESim Ansys Solver: Exchange FEA data with Ansys and execute the solver
22 Ex: Nastran Component Editor Input tab defines input bulk data file Filter by load, element property, material, etc. Select which input parameters to expose to isight
23 Nastran Component Editor Execution tab defines Nastran command line and other runtime behavior User can configure which parameters are supported through a configuration file User can control how INCLUDE files are handled
24 Ex:Nastran Component Editor Output tab defines output f06 file to be parsed Select which output parameters to expose to isight
25 Solution Components Suite CAE Performance Simulation Ansys Workbench: Exchange FEA data with Ansys Workbench and execute Ansys solver AVL: Exchange data with and execute AVL solvers Femap: Exchange FEA data with Femap, execute Nastran and other solvers Flotherm: Exchange data with Flotherm CFD GT-Power Engine Combustion: Exchange data with and execute GT-power engine simulation software LS-DYNA/LS-OPT: Exchange data with and execute LS-DYNA through LS-OPT
26 Solution Components Suite CAE Performance Simulation Madymo Occupant Safety: Exchange data with Madymo generic multibody and FEA software Nastran: Exchange FEA data with Nastran and execute the solver Patran: Exchange geometry and FEA data with Patran, launch Nastran, Marc, and other solvers SimulationX: Exchange data with and execute ITI SimulationX STAR-CCM+: Exchange data with and execute STAR-CCM+ CFD
27 Solution Components Suite CAD Geometry Modification CATIA V5 CAD Driver: Exchange CAD data with CATIA V5 CATIA V5 CAT_Analysis: Exchange CAD and simulation data with CATIA embedded analysis tools Pro/E CAD Driver: Exchange CAD data with ProEngineer Unigraphics CAD Driver: Exchange CAD data with Unigraphics Geometry to CAE 2Abaqus: Create an Abaqus input deck from CAD geometry SFE Concept: SFE Concept
28 Design Drivers Task Plan Optimization Component DOE Component MCS Component Approximation Model Visual Design Driver
29 29 Design Study Tools Central Composite Orthogonal Array Latin Hypercube Full Factorial Parameter Data file Optimal LHC Design of Experiment Approx. Models Taylor series Response Surface Stepwise RSM (9.0) Variable complexity Kriging RBF Quality Engineering Monte Carlo Taguchi Robust Reliability Analysis Reliability Optimization Reliability based Robust Design Trade-off Analysis Optimization Multi-objective GAs Engineering Data Mining Rule-based (2) Exploratory (3) Gradient-based (7) Mixed Variable (2) Pointer Stochastic Design Improvement (SDI)
30 Task Plan Create Task Plan to execute combinations of multiple design drivers Equivalent to isight Task Plan Task Plan GUI
31 Optimization Component Some optimization components available NLPQL, Hooke-Jeeves, LSGRG2 Multi-island genetic algorithm (MIGA) Pointer automatic optimizer Multi-objective genetic algorithms (NSGA-II, NCGA) Pointer Component Editor GUI NSGA-II Component Editor GUI
32 Design Of Experiments Component DOE component methods available in v1.0 Central Composite Data File Full Factorial Latin Hypercube Optimal Latin Hypercube Orthogonal Array Parameter Study DOE Component Editor GUI
33 Monte Carlo Simulation Component Monte Carlo Simulation component Sampling methods Simple Random Sampling Descriptive Sampling Monte Carlo Component Editor GUI
Six Sigma Component (cont.) General Setup Run mode Analysis Optimization Classes of plugins Reliability Techniques Monte Carlo Sampling DOE Description Execution Options 34
Six Sigma Component Random Variables and Responses tabs same as Monte Carlo New Discrete- Uniform distribution 35
Six Sigma Component (cont.) Six Sigma Results parameter Select outputs to be included in aggregate parameter Allows mapping to other component to drive designing for uncertainty quality improvement process 36
Six Sigma Component (cont.) Optimization configured within Six Sigma component editor Robust optimization option creates model with six sigma component embedded in optimization component 37
Six Sigma Results Six Sigma Component methods/results include: Six Sigma Analysis Sigma Level / Quality Level Defects per million Reliability Analysis Reliability (probability of success) Probability of failure Statistics Mean, Standard deviation, etc. Robust / Reliability Optimization New robust/reliable design point 38
Taguchi Component P-Diagram summary of setup Control Factors, Array Noise Factors, Array Signal Factor, number of levels Responses System type defined up front: Static Dynamic 39
Taguchi Component (cont.) Response Definition Customized for static or dynamic system Type: Lower is better Nominal is best Higher is better Dynamic quality characteristic Zero point proportional Reference point proportional General linear equation Target Loss constant Static Dynamic 40
Taguchi Robust Design Results Main Effects Graph Dynamic Response Graph 41
Taguchi Robust Design Results (cont.) Taguchi Results Table Static Dynamic Taguchi Response Summary 42
Interactive Main Effects Viewer Display/Select: Best levels Worst levels Best point Worst point Baseline point Filter displayed factors Select individual levels, S/N and Sensitivity estimated Benefit over baseline of selected point Execute noise/signal runs to confirm estimates 43
Correlated Random Variables Correlation Matrix button Correlation Matrix editor table Correlation results table 44
45 Approximation Model Component Approximation models available in v1.0 Response Surface Model (RSM) Radial Basis Function (RBF) neural network approximation Approximation model capabilities Wizards for setting up the model Inspection tools for evaluating model precision and margin Visual Design Driver 3D approximation model viewer for surfing the design space Approximation models can be created as independent activities or applied to any subflow or activity component
46 Inspection Tool for Approximation Models Special GUI for approximation model inspection Evaluate precision Show margins between each response value Show present value and predictive value
47 Visual Design Driver Visualize and surf the approximation model design space Highly interactive, multi-dimensional GUI Change evaluation values by moving slide bars with mouse Displays multiple thumbnail plots, any of which can be switched into the main window Main 3D model Parameter panel Thumbnail plots
48 Workflow Execution & Post Processing
49 Monitoring Tools Runtime Workflow status Monitoring with graphs
50 POST Processing and History Tables Post processing for sampling results Execution history to check values
51 Engineering Data Mining for Post Processing Main GUI Scatter Plots
Data Analysis Tab The Runtime Gateway contains a single Data Analysis tab with a number of useful post processing tools 52 52
Data Analysis Tab Including EDM 53 53
Data Analysis Tab Correlation Map 54 54
Data Analysis Tab And a Grid of 2D Scatter Plots 55 55
Correlation Map Tab Where selecting a parameters is synced with the rest of the Gateway 56 56
Correlation Map Tab Where selecting a parameters is synced with the rest of the Gateway 57 57
Correlation Map Tab The old functionality remains (though moved around a bit) Correlation values now displayed as a tooltip 58 58
59 Extending Local Processes to the Enterprise
60 xtensible to the Enterprise Publish Components and Workflows to FIPER Internet LAN, WAN, VPN Desktops Clusters Servers FIPER ACS (Application Control Server) Connection Share Components and Workflows Effective use of computer resources on FIPER Monitoring executing Workflow status on Web Consolidate data management using Database
61 System Simulation Challenges Simulation of systems often require multiple simulation models, significant data management, and workflow definition. This is often performed in a stove pipe fashion with communication in manual steps. As a result, design alternatives and important trade-offs and go un-noticed and are never communicated to the customer.
62 Simulation Collaboration Challenge Design groups require and request simulation results from other groups. Design time is lost while waiting for response. As a result, this limits and delays options.
FIPER Distributed and Parallel Computing Leverage parallel and distributed computing for efficient multirun simulation isight framework is parallel and distributed by design IBM On-Demand "Grid computing" environment Transparent to end-user No shared file system requirements Any mix of operating systems Any mix of hardware platforms No special configuration files or model changes No limitations on structure of model Open to leverage existing grid environment: E.g. LSF, Globus, etc. Heterogeneous Compute Environment No 3 rd party grid software required 63
64 FIPER - Collaborative Design FIPER allows knowledge sharing and design collaboration by: Provides a model and tool library Remote process sharing and execution Incorporate other's processes in your own Real-time data access Protects intellectual property Shares best practices Securely collaborate with coworkers, vendors, or partners Internally (LAN) Externally (B2B internet) Prime contractors Turbine Process Sub contractor Blade Process Sub contractor Disk Process
65 FIPER WEBTOP Client FIPER allows end user to access, modify, and execute models from a web client. FIPER allows users to run model remotely through browser Web server
66 FIPER and Simulation Data Management Data Data anywhere Simulation Data Data Management Simulation Process Process Management Decision Support Support Unmanaged files Traceability Automation Design to target + Managed Processes + Managed Performance Attributes Simulation driven performance attributes Islands of unconnected data Files + Managed Files + Metadata Program & non program data Standard & ad-hoc Product System Dashboard Target k Target l Target m Simulation Marketing Requirements Target Performance Attributes
67 isight-fd Component Central Demo Videos and the latest additions to the components suite are available at: http://components.engineous.com
68 Thank you!